Performance analysis of models used to predict failure in secondary school

Abstract

According to data made available by Pordata for the year 2021, about 8.3% of medium school students in Portuguese schools fail or drop out of the educational system, another 9.8% of students in this situation are still in basic education. Since education is one of the pillars of a country’s development, it is important to understand the reasons behind these statistics and discover what leads students to such failure in order to try to mitigate these results. In order to do so, it is necessary to acquire data about the students, thus emerging the area of Educational Data Mining. Early prediction of school failure can be the key piece of the effort to avoid it. So, this paper present a comparative study of machine learning models to indicate the best model to predict school failure.